On 11/10/12 00:45, Marek Vasut wrote: > Dear Gerd Hoffmann, > >> On 11/09/12 10:50, Peter Maydell wrote: >>> On 9 November 2012 10:42, Anthony Liguori <aligu...@us.ibm.com> wrote: >>>> While the abstract discussion is fun, it never hurts to be defensive. I >>>> agree the root cause is vmware-vga but checking in vnc doesn't hurt. >>> >>> Defensive programming would suggest doing the clipping in the >>> console.c layer. That sounds a reasonable plan to me (especially >>> if we've hit similar problems multiple times in the past). >> >> Fully agree, I'll cook up a patch as I'm touching that anyway. >> >> Question is just whenever we'll go silently fixup stuff in console.c or >> use assert()s to enforce callers getting this correct. I'd tend to use >> assert() as vmware-vga passing bogous stuff there IMHO indicates there >> is a bug in vmware-vga. > > Or rather some revisions of the guest X driver. Though it's worth > investigating > it in the right place indeed.
That too, but we must add a check to qemu nevertheless. We can't trust the guest to not pass in bogous data, be it intentionally or by mistake. vmware-vga must sanity check the guest input no matter what, but validating the guests input once should be enougth. cheers, Gerd
